CMOS image sensors have been widely used in various products such as cell phones, tablet computers, security monitoring systems and the like. In many applications such as industrial camera, machine vision and the like, it is required to capture the objects moving at high speed without distortion, due to the high speed motion of the object and the requirements of image recognition algorithm. Traditional CMOS image sensors with rolling shutter, which read progressively, may generate Jell-O effect on objects moving at high speed and the image may be distorted, and therefore, it is required to use a global shutter.
However, in the existing image sensors with a global shutter, the two read-outs of the image signal and the reset signal are not correlated in a full sense in which the signal is read out before the signal is reset, and thus the noise in the signal cannot be entirely eliminated, which leads to decreased SNR of image. Currently, no solutions are available to address this technical problem.